We use a hybrid type-2 pragmatic effectiveness-implementation study design and a parallel cluster RCT design (Figure 1) to evaluate the AA-SH and AA-MAP implementation strategies for smoking cessation at 8 ART clinics in the 6 most populous provinces or regions across Laos. We used the CONSORT (Consolidated Standards of Reporting Trials) [48] and SPIRIT (Standard Protocol Items: Recommendations for Interventional Trials) [49] checklists in the preparation of this paper.

Several aspects of this project were carefully considered before deciding on the proposed design. We chose a cluster (vs individual) RCT design as it is a robust, pragmatic design for service interventions (AA-SH and AA-MAP) at the cluster (ART clinics) level rather than the patient level. We considered a stepped-wedge design, which would have allowed all clinics to receive both interventions during this trial and within-cluster comparisons. However, the drawbacks of this design are the risk of unequal chronological exposure to interventions, and the potential risk of internal contamination due to carry-over effects (ie, during the treatment period, enrollees may be exposed to the other intervention if the clinic switched interventions). Thus, we chose the parallel cluster RCT design but will allow all clinics the option to implement both AA-SH and AA-MAP after this trial’s period.

To guide our implementation work, we have been using and will continue to use the PRISM [45-47]. We selected PRISM because it includes the RE-AIM framework, which helps to measure the desired outcomes of AA-SH and AA-MAP (ie, reach and effectiveness) in our previous studies and in aim 1 of this study. PRISM expands RE-AIM to include more key, multilevel contextual factors relevant to program implementation and sustainment, particularly in adopting technology-based interventions in clinical settings.

We use the phase-based model (PBM), a theoretical framework that is specific to smoking cessation [50], to guide our MAP treatment. PBM partitions the cessation process into 4 phases: motivation, preparation (precessation), cessation (quit date to 2 weeks post quit), and maintenance (up to 6 months post quit). PBM helps to identify challenges or opportunities at each phase, explains underlying phase-specific mechanisms, and facilitates selection of intervention components and measures. Among several putative mechanisms, some are relevant across phases and have been empirically shown to be reliably associated with long-term abstinence. These include withdrawal and craving, motivation to quit, positive or negative affect, coping with stress and urges, self-efficacy, and perceived support [50-54]. As such, MAP specifically targets these mechanisms.

It is estimated that Laos had 17,000 people with HIV in 2022 (female=38%) [55]. The number of new cases were 1550 in 2021 and 1748 in 2022 [31]. Of those diagnosed, 97% of them are registered with an ART clinic, and 77% of them are on ART. Of those on ART, 96% of patients are aged ≥18 years, and the mean duration on ART is 6.6 years [31]. Registered patients have periodic prescheduled appointments (about once per month for new patients and once per 4 months for patients with controlled viral load).

Of the 11 main ART clinics in Laos, CHAS recommends implementing this project at 8 ART clinics in the 6 provinces or regions that are most populous or have an increasing HIV+ incidence. Besides currently registered patients, these clinics together gain approximately 1500 new patients each year [31]. These clinics provide comprehensive care for registered ART patients, including ART, ART adherence counseling, nutrition, risk-reduction counseling, and treatments for comorbidities other than tobacco use. To avoid potential skewness, we created 2 sets of 4 clinics with comparable patient populations and available resources and then randomly assigned each set to an implementation strategy (AA-SH or AA-MAP).

All potential implementers will receive an initial 2-hour in-person group training on implementing Ask-Advice-Connect and collecting data using REDCap. The training also includes information on why people with HIV should quit smoking, the effectiveness of the Ask-Advice-Connect approach in high-income countries, health care provider roles, available nicotine replacement therapies (NRTs) in Laos, and time for questions and answers. The initial training will be held immediately before the launch of the AA-SH or AA-MAP implementation at the ART clinics and will emphasize the importance of fidelity to the interventions. The initial training will be video recorded for future review and for the training of newly hired clinic staff. There will be 4 “booster” training sessions at months 1, 6, 12, and 18 to reinforce the information in the initial training, address emerging issues, and answer clinic staff’s questions. These 30-minute booster sessions will coincide with required clinic-wide staff meetings. Both the initial and booster training sessions will involve didactic lectures and role-playing exercises.

Inclusion criteria for patient participants (ie, people with HIV who smoke) in the cluster RCT for the assessment at 6 months are the following: (1) being aged ≥18 years; (2) reachable within 30 days of the target assessment date by different methods (up to 4 phone calls on different days and times, 3 text messages for each platform [eg, SMS, Telegram, or WhatsApp], 3 emails if applicable, and via direct contact at a prescheduled clinic appointment if applicable), (3) consenting (at the time of enrollment) to be contacted at 6 months post enrollment, and (4) able to provide a written informed consent at month 6 when contacted by the research staff to participate.

Participants and inclusion criteria for other study components are the following: (1) being aged ≥18 y, (2) employed by the clinic or institution for at least 3 months, and (3) able to provide written informed consent to participate (Table 1).

There were no exclusion criteria for this protocol.

People with HIV who smoke and are willing to make a quit attempt within the next 2 weeks will be offered treatment (self-help guide or MAP). Reach is the proportion of those who enroll in treatment. There are >16,000 registered patients at the 8 ART clinics. Given the male and female ratio and estimated smoking prevalence by sex, we estimate that 5760 male people with HIV and 450 female people with HIV are current smokers. Our previous and ongoing studies in Cambodia show that 65%‐80% of people with HIV who smoke are willing to make a quit attempt and enroll in treatment [34]. Conservatively assuming that only 50% (3105/6210) of Lao people with HIV who smoke agree to be enrolled in AA-SH or AA-MAP, there will be at least 2880 male and at least 225 female participants. Of these, based on our previous studies and the rate of smartphone ownership among Lao people with HIV, we predict that the reach of AA-SH will be 90% and the reach of AA-MAP will be 70%.

Participants in the AA-MAP arm are asked to answer some questions on day 1 and to complete brief weekly assessments (4‐6 questions) via the Insight app on smartphones for 6 months. Responses to these questions are used to personalize and tailor MAP treatment to each participant.

The pool of participants for this assessment is enrollees in AA-SH and AA-MAP. At 6 months post enrollment, a local research team member (not associated with ART clinics) will call the enrollees to assess smoking status. Participants who consented to being contacted after 6 months at enrollment and are reachable within 30 days of the target assessment date will undergo an informed consent process conducted by the research staff. Participants who provide written informed consent will be asked to complete a 30-minute survey in person at the ART clinic or via a phone or video call. Data will be collected using the REDCap mobile app on a tablet [63,64].

The total implementation period will be up to 36 months (enrollment will stop at 30 months and the last enrolled participant will be assessed at 36 months). We will assess up to 1200 treatment enrollees at 6 months. In our previous and current Lao and Cambodian studies, retention was 98% at 3 months and 90% at 6 months. Conservatively assuming only 60% (1863/3105) of AA-SH and AA-MAP enrollees can be contacted and agree to participate, there will be at least 1728 male and 135 female enrollees available to complete the 6-month assessment. Thus, a sample of 1200 enrollees is highly feasible. Given the hypothesized greater reach of AA-SH (90%) over AA-MAP (70%), we anticipate around 700 in the AA-SH group and around 500 in the AA-MAP group.

The primary outcome to evaluate effectiveness is smoking status at 6 months post enrollment. Participants who self-report 7-day point prevalence abstinence will be seen in person at the clinic within 30 days post survey for biochemical verification. We will try to arrange this visit to coincide with participants’ regular or prescheduled clinic visit for ART or other services. At the in-clinic appointment, they will be asked again, “Did you smoke cigarettes in the past 7 days?” before breath carbon monoxide will be measured by Smokerlyzer [65]. Abstinence is defined as having expired carbon monoxide <6 ppm [66].

We will also estimate the impact of AA-SH and AA-MAP. Impact, the multiplicative effect of reach and effectiveness, is crucial in evaluating health intervention implementation. An intervention may have a broad reach but limited effectiveness, or have strong effectiveness but limited reach, and hence ultimately may have limited public health impact [67,68]. Other measures of aim 1 are listed in Table 2. Paper and District Health Information Software 2 [69] medical records will be used to collect patients’ clinical information (with their consent), such as most recent CD4 lymphocyte counts, viral loads, medication, tuberculosis, and other comorbidities.

The primary outcome for reach is the percentage of smokers willing to make a quit attempt who accept the self-help guide (in the AA-SH condition) or sign up for the MAP (in the AA-MAP condition). We hypothesized that reach would be greater for AA-SH (90%) than for AA-MAP (70%). Generalized linear mixed models (GLMMs) will be applied to evaluate the difference between AA-SH and AA-MAP. In addition to condition as a fixed effect, the primary model will include clinic as a random effect, given the potential for differences in acceptance and downloading due to the clinic. Expected denominators are at least 3000 (2880 men and 225 women) for both AA-SH and AA-MAP. With α=.05, a 2-sided test, an intracluster correlation coefficient of 0.22 or less, and the hypothesized enrollment rates of 90% for AA-SH and 70% for AA-MAP, power is greater than 0.99. Furthermore, power will be greater than 0.80 for enrollment rates of AA-MAP that are at least 5% lower than AA-SH with AA-SH rates of 60% or higher. Additional analyses will explore factors associated with reach within the AA-SH and AA-MAP conditions to guide future adaptation of either intervention.

The primary outcome for effectiveness is the biochemically verified 7-day point prevalence abstinence at the 6-month assessment. Those who fail to complete biochemical verification will be imputed as smoking. Based on our previous studies with US or Cambodian people with HIV, we predict abstinence rates at 6 months will be 8% (4/50) for AA-SH and 16% (8/50) for AA-MAP (odds ratio [OR] 2.20). This estimate of abstinence in the MAP group is based on the high end-of-treatment abstinence (40%) observed in our pilot RCT with Cambodian people with HIV, along with the many observations of high relapse rates at long-term follow-up among people with HIV [89,90]. GLMM will be used to evaluate the difference between AA-MAP and AA-SH in abstinence rates at 6 months post enrollment. Condition will be entered as a fixed effect in the model, along with clinic as a random effect. Any smoking-related or demographic variable found to differ with P<.10 across the conditions will also be included in the model (eg, biological sex). With an expected 700 participants in AA-SH and 500 in AA-MAP, α=.05, a 2-sided test, and an intracluster correlation coefficient of 0.10, power is greater than 0.98 to detect the predicted difference between the intervention groups. Furthermore, power will be greater than 0.80 for (1) abstinence rates equal to 8% for AA-SH and 13% or higher for AA-MAP (OR 1.72) with the same sample sizes, or (2) for abstinence rates equal to 8% and 16% (OR 2.20) with sample sizes of 300 for AA-SH and 250 for AA-MAP. Exploratory analyses will be performed. Differences between those who are abstinent and those who are not abstinent will be evaluated using t test and a chi-square test. GLMMs or linear mixed models will be used to examine other smoking-related outcomes, such as self-reported continuous abstinence, prolonged abstinence, and number of quit attempts. Additional analyses will explore factors associated with effectiveness within the AA-SH and AA-MAP conditions to guide future adaptation of either intervention.

For quantitative data (staff survey and REDCap data), we will use descriptive statistics to characterize clinics, staff, and implementation outcomes. We will also use appropriate statistical methods (eg, chi-squared test or logistic regression) to examine the associations between clinic or staff characteristics and implementation outcomes. All qualitative interviews (IDIs and KIIs) will be recorded, transcribed verbatim, and then translated into English for analysis using thematic analysis [91]. We will use a hybrid approach [92] for the thematic analyses by developing deductive codes [93] based on PBM and PRISM constructs and inductive codes that emerge from the data [94]. At least 2 research team members will code all transcripts with the aid of MAXQDA [95]. Discrepancies will be discussed to achieve intercoder agreement. We will ask a small group of interview participants (n=5) to review the themes to ensure that our interpretation of the data is consistent with their experiences (ie, member checking) [96]. We will develop a mixed methods data inventory [97] and a matrix to examine quantitative and qualitative data jointly. We will use the integration strategies of merging and corroborating with the intent to explore what might have affected implementation outcomes [97,98]. We will also use data reduction to identify data that were primarily useful for this analytical focus (ie, qualitative and quantitative data of the same themes or areas but may converge or diverge). Mixed methods linked data, results, and meta-inferences will be presented in visual joint displays [98], similar to our previous publication [34,99].

Information on the resources needed, total costs, and cost-effectiveness estimates of AA-SH and AA-MAP are critical for partnering with Lao government agencies to implement and scale up the interventions at ART and other clinics throughout Laos. Therefore, to inform decision-making of public sector stakeholders, we will conduct an analysis of the resources needed to widely implement the interventions, and the potential gain from investing those resources from the perspective of the government service providers.

Our resource use and effectiveness methodology will follow current best-practice guidelines [100-102]. In keeping with the PRISM framework and the goal to inform the Lao government of resource use, costs, and gains from AA-SH or AA-MAP implementation, the analysis of resources will be from the perspective of government payers, initially not including related expenses such as patient out-of-pocket costs. In addition, we will use the gold-standard methodology of measuring resource use and assigning generalized prices to each type of resource used [103-105], resulting in “costs” that are comparable across different intervention sites and patient populations. We have successfully used this methodology in a variety of previous studies [106-111].

We will develop a project-specific version of the cost assessment methodology (CAM) template as we did in previous studies [106-111]. The CAM template streamlines the process of identifying and collecting data on the resource usage and costs of community and clinical interventions. Advantages of the CAM approach are that the cost results are not tied to the specific location of the intervention, which may have higher or lower general costs (eg, city vs rural), and it does not “miss” intervention resources—and thus costs—that may have been donated or subsidized for this study. Consequently, the data collected as part of aim 3 can be used to predict the costs for implementing the interventions in new areas and/or countrywide in Laos. The project-specific data for the CAM template will be informed by steps 1‐3 listed below, and calculations in step 4 will use these data. All data delineation and collection will be completed in close consultation with all of the investigators, particularly with the Lao coinvestigators.

(1) Identify the individual components and phases of the interventions, and the resources needed to conduct each component. Examples of resource categories include personnel effort (eg, clinic staff and the national MAP core), material items (eg, self-help guides), and MAP infrastructure (eg, data storage). (2) As closely as possible, track the number of “units” of each type of resource that are used over the course of the implementation of each of the 2 intervention strategies (eg, hours of personnel time, number of self-help guides delivered, and MAP server rentals). Where precise or even approximate tracking of specific inputs to the intervention is not possible, the research team will develop estimates, including a range that can be used for sensitivity analyses. (3) Identify a generalized “price” per unit for each type of resource to determine a total cost for that resource by multiplying price by the total number of resources. These prices will be drawn from Lao national or representative sources. For example, the price per hour for the counselors or nurses will be collected from the Lao MOH payroll record. (4) Calculate total units and costs of the different resources used; total costs per intervention strategy; cost per patient participant in each RCT arm; and cost-effectiveness of the AA-MAP relative to AA-SH using the abstinence rate in each arm at 6 months.